The evaluation of your submission will be based on the predictive R squared between the predicted response and the true response, calculated as follows:

$$ \text{MSE} = \frac{1}{n} \sum_{i=1}^{n} (y_i - \hat{y}_i)^2 $$

Where \( y_i \) is the true response value and \( \hat{y}_i \) is your predicted value for observation \( i \). And:

$$ R^2 = 1 - \frac{\text{MSE}}{\mathbb{V}(y)}$$

Your rank on the leaderboard will be determined by:

$$ R^2 $$

But your grade will be determined by:

$$
\text{Score Student i} = \frac{R^2_i}{\text{max } R^2_i}
$$

Where:

$$ max R^2_i $$

is the maximum R squared in the leaderboard.

The closer your R squared is to the maximum R squared, the higher your ranking on the leaderboard. This means that you will need to find the most effective model that maximize your R squared, while exploring innovative techniques to uncover the hidden structure within the data.

Good luck!